update_ahmia_blacklist.py 1.0 KB

12345678910111213141516171819202122232425262728293031323334353637
  1. #!/usr/bin/env python
  2. # lint: pylint
  3. # SPDX-License-Identifier: AGPL-3.0-or-later
  4. """This script saves `Ahmia's blacklist`_ for onion sites.
  5. Output file: :origin:`searx/data/ahmia_blacklist.txt` (:origin:`CI Update data
  6. ... <.github/workflows/data-update.yml>`).
  7. .. _Ahmia's blacklist: https://ahmia.fi/blacklist/
  8. """
  9. # pylint: disable=use-dict-literal
  10. from os.path import join
  11. import requests
  12. from searx import searx_dir
  13. URL = 'https://ahmia.fi/blacklist/banned/'
  14. def fetch_ahmia_blacklist():
  15. resp = requests.get(URL, timeout=3.0)
  16. if resp.status_code != 200:
  17. # pylint: disable=broad-exception-raised
  18. raise Exception("Error fetching Ahmia blacklist, HTTP code " + resp.status_code)
  19. return resp.text.split()
  20. def get_ahmia_blacklist_filename():
  21. return join(join(searx_dir, "data"), "ahmia_blacklist.txt")
  22. if __name__ == '__main__':
  23. blacklist = fetch_ahmia_blacklist()
  24. with open(get_ahmia_blacklist_filename(), "w", encoding='utf-8') as f:
  25. f.write('\n'.join(blacklist))